I wrote a Solr service for Whirr a few months ago against 0.4.0 and am updating it to 0.6.0 (I need to see about getting this committed). In looking at the new internal Whirr API changes from 0.4.0 to 0.5.0 I noticed the new start and stop scripts. And that got me thinking about writing custom teardown logic.
For example, when I destroy a cluster, it would be nice if the before / afterDestroy methods were called, and allowed the service to put together a script that gracefully stops the service (finish processing any current requests, but refuse new requests), or backup of any state I wanted to store (maybe to S3 or EBS). I was hoping that this is how it was working in 0.6.0. By the way...still love whirr! On Wed, Aug 31, 2011 at 10:02 AM, Andrei Savu <[email protected]> wrote: > On Wed, Aug 31, 2011 at 9:15 AM, John Conwell <[email protected]> wrote: > > When the user calls 'whirr destroy-cluster' or 'whirr destroy-instance' > from > > the cmd line, does whirr execute the stop scirpt for the services defined > in > > the config file? > > No. We've added the stop scripts for future use (e.g. add / remove > node functionality). > > > Also, is this when the ClusterActionHandlerSupport.beforeDestroy > > and afterDestroy methods get called? > > before / afterDestroy are actually never called right now. IMO this is > a bug we need to fix. How are you planning to use this event? > -- Thanks, John C
